Date: Wed, 10 Dec 2014 15:33:25 -0500 From: John Baldwin <jhb@freebsd.org> To: "Christopher R. Bowman" <crb@chrisbowman.com> Cc: "freebsd-hackers@freebsd.org" <freebsd-hackers@freebsd.org>, Adrian Chadd <adrian@freebsd.org>, Tony Maher <tonymaher@optusnet.com.au> Subject: Re: 10.1 hangs during boot on Intel DP35DP board Message-ID: <201412101533.25855.jhb@freebsd.org> In-Reply-To: <A2E116B4-1A45-4059-89B8-912A994ADD52@ChrisBowman.com> References: <1dbd10028e6466adac4d5c10cf7e099d4fe035c1@webmailnew.optuszoo.com.au> <3125619.Pmuf3ojzUs@ralph.baldwin.cx> <A2E116B4-1A45-4059-89B8-912A994ADD52@ChrisBowman.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tuesday, December 09, 2014 5:17:47 am Christopher R. Bowman wrote: >=20 > > On Dec 8, 2014, at 10:32 AM, John Baldwin <jhb@freebsd.org> wrote: > >=20 > > On Saturday, December 06, 2014 01:37:09 PM Christopher R. Bowman wrote: > >> John, > >> Its still sitting but I turned on verbose output and it's been sitting= =20 for > >> 10 minutes so far without any further output. Christopher. > >=20 > > Grrr, then it might be stuck elsewhere than the interrupt config hooks.= =20 > > Unfortunately we'd really need to find a way to get into DDB. I think= =20 GENERIC=20 > > doesn't include BREAK_TO_DEBUGGER by default, so Ctrl-Alt-Esc doesn't w= ork=20 out=20 > > of the box. Can you break into the loader prompt and type 'set=20 > > debug.kdb.break_to_debugger=3D1' before booting? Ctrl-Alt-Esc should t= hen=20 work=20 > > when it hangs. Once that happens, please get the output of 'ps' from D= DB. > >=20 > >> -- > >> Christopher R. Bowman > >>=20 > >>> On Dec 5, 2014, at 7:20 AM, John Baldwin <jhb@freebsd.org> wrote: > >>>=20 > >>> On Thursday, December 04, 2014 07:42:48 PM Christopher R. Bowman wrot= e: > >>>>> On Dec 4, 2014, at 8:39 AM, John Baldwin <jhb@freebsd.org> wrote: > >>>>>> On Wednesday, December 03, 2014 9:36:01 am Adrian Chadd wrote: > >>>>>> Hi, > >>>>>>=20 > >>>>>> I'm not sure - it may be that the timekeeping stuff is a bit=20 confused. > >>>>>> Maybe it's now time to file a bug and see if we can loop the > >>>>>> timekeeping folk into it. > >>>>>=20 > >>>>> This is probably not timekeeping. That message is just the last th= ing > >>>>> printed before the interrupt-driven hooks are run. If you booted a= =20 HEAD > >>>>> snapshot the kernel should have DDB compiled in. Please drop into = DDB > >>>>> using Ctrl-Alt-Esc and then run 'show conifhk' and reply here with = the > >>>>> output please. > >>>>=20 > >>>> John, > >>>>=20 > >>>> Thank you for responding. I tried that but the machine seemed wed= ged > >>>> at > >>>>=20 > >>>> that point. I can=E2=80=99t drop into DDB using Ctrl-Alt-Esc nor ca= n I do > >>>> Ctrl-Alt-Del to reboot. All I can do it push the reset button. > >>>> Christopher > >>>=20 > >>> If you let it sit for 60 seconds it should print out a message tellin= g=20 you > >>> what it is waiting on. > >=20 > >=20 > > --=20 > > John Baldwin >=20 >=20 > John, > I tried this twice on the off chance I typed it wrong once. This had no= =20 effect. After letting the machine sit for 5 minutes there was no further=20 output or prompt. Doesn=E2=80=99t look like it dropped into the debugger. = Any more=20 suggestion? Can you trigger an NMI? Perhaps remotely via IPMI? (I'm not sure if that'= s=20 even feasible in the IPMI spec.) Barring break_to_debugger, that's the onl= y=20 other thing I'm aware of. Another option might be to compile a custom kern= el=20 with 'options KTR', 'options KTR_VERBOSE', 'options KTR_COMPILE=3DKTR_PROC'= and=20 'options KTR_MASK=3DKTR_PROC' and boot that. If the machine is truly hung = it=20 will still hang. If you can point the machine at a serial console and log = the=20 output, that would be good if it doesn't really hang but is sitting idle. =2D-=20 John Baldwin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201412101533.25855.jhb>